New printer design with new 'twist' for XY on Z

I came up with this idea about a year ago seeing an industrial wood planing machine use a similar approach, and I thought it could make for a neat and great printer design, but before ordering the expensive parts, I thought I'd post here first to see if there would be any issues with physics/mechanics.

Basically, the machine is a XY platform moving on the Z axis, with a new 'twist'; there are two leadscrews fixed diagonally from each other to the base, with the leadnuts rotating around the leadscrew. The leadscrews are M22x5mm pitch, single start. The two red parts are a high temp plastic flanged bushing that the brass leadnuts rotate within.

In the opposite two corners, there are 20mm linear shafts with corresponding linear bearings.

Both the leadscrews and the linear shafts will have the ends turned down and will be press fitted into the printer's bottom plate.

All of the parts will come from Misumi. I didn't model the actual printhead mechanics, as I didn't want to dive too far into that yet.

Bottom plate and XY plate will be 1/2 in. thick Mic-6 aluminum. Approximate printer physical dimensions: 500mm^3. The heated bed will be a seperate Mic-6 plate with silicone heater adhered to the bottom and will be isolated from the printer's bottom plate with ceramic standoffs.

The printer will be easy to make an enclosure for, and the linear shafts are the hollow variant, so cables can be routed cleanly through them.

One issue I see is the two leadnuts becoming unsynchronized perhaps, but this may not happen often. This could be solved with a single motor and a closed loop belt, if you can get a belt that long.

I really like this design! I am looking for positive feedback. Thank you!

Render:


Edited 17 time(s). Last edit at 03/04/2018 02:13AM by klcjr89.

Interesting!
I'd definitely add a fixed top frame with cross braces, because the stability when the printhead is down, is pretty low.

Quote
o_lampe
Interesting!
I'd definitely add a fixed top frame with cross braces, because the stability when the printhead is down, is pretty low.

Are you saying to add a 1/2" top plate and connect this top plate to the bottom plate with other plates/diagonal pieces? Using plates would serve double duty as the enclosure. I would think I could get away with putting a 1/2" plate on top and then for the sides of the cube, use Lexan to make the enclosure?

Also, do you see any problems with print layer registration or the XY plane moving around while the Z axis moves up? The holes will be cnc machined to ensure perpendicularity and correct spacing for orthogonality.

I hope the minor deviation in the linear shaft and leadscrew straightness won't be an issue? The 500mm leadscrew has a runout spec of 160 microns or less, and the 500mm linear shaft straightness spec is 50 microns or less.

Thanks!

Edited 8 time(s). Last edit at 03/04/2018 05:01PM by klcjr89.

You need some way to keep the shafts vertical, pressing them in wont provide enough resistance to side loads and your printer will fold like a parallelogram. The design is very space efficient, but supporting your linear shafts on only one side will cost a lot of rigidity. TBH the math doesn't look great even with 20mm shafts.

Quote
klcjr89
Are you saying to add a 1/2" top plate and connect this top plate to the bottom plate with other plates/diagonal pieces?

Also, do you see any problems with print layer registration or the XY plane moving around while the Z axis moves up? The holes will be cnc machined to ensure perpendicularity and correct spacing for orthogonality.

I hope the minor deviation in the linear shaft and leadscrew straightness won't be an issue? The 500mm leadscrew has a runout spec of 160 microns or less, and the 500mm linear shaft straightness spec is 50 microns or less.

A top plate will make the design perform much better, but I would still be a little worried about the XY stage tilting diagonally if you are only using one row of linear bearings.

The interaction between the screws and the linear shafts won't be an issue since a hundred microns or so of divergence will get absorbed by bending in the shafts/frame, but you need to think about which degrees of freedom are being constrained.

It is quite likely that the leadscrew nuts will have enough radial play that the linear shafts will provide all the XY registration and the screws will only handle the Z load. If that is the case, your 22mm leadscrews are doing almost nothing. I would be very carefully reading datasheets and maybe going with rolled ballscrews instead.

Edited 1 time(s). Last edit at 03/04/2018 08:33PM by 691175002.

Quote
691175002
You need some way to keep the shafts vertical, pressing them in wont provide enough resistance to side loads and your printer will fold like a parallelogram.

The design is very space efficient, but supporting your linear shafts on only one side will cost a lot of rigidity. TBH the math doesn't look great even with 20mm shafts.

What side loads would there be? It's a printer, so any side loads should be negligible?

Could you elaborate on the math part. I'm more of a visual person, so any kind of picture would help.

Thank you!

Edited 2 time(s). Last edit at 03/04/2018 08:24PM by klcjr89.

I edited my post with a few more details. The side-loads on a 3d printer might be small, but you are still looking at maybe 200-400g of acceleration depending on the speed and extruder being used.

If you put that load on the very end of a 500mm rod that is only being held by a tiny section on one end it will still bend quite a bit. For example here is an extremely rough FEA where I modeled the whole assembly as a solid block of steel and it moves 0.05mm when 200g of force is applied. Its not awful, but the simulation is probably 2-4 times more rigid than your actual build will be and that kind of deflection will be visible.



The bigger problem IMO is that if you leave the rods unsupported and someone bumps into them you are essentially applying half a meter of leverage to a very small hole in a piece of cast aluminum. I'd guess 2-5Kg at the top of the rod could permanently damage the alignment of the printer.

Again, not awful, but not really something I'd want to risk.

Edited 2 time(s). Last edit at 03/04/2018 08:58PM by 691175002.

Quote
691175002
I edited my post with a few more details. The side-loads on a 3d printer might be small, but you are still looking at maybe 200-400g of acceleration depending on the speed and extruder being used.

If you put that load on the very end of a 500mm rod that is only being held by a tiny section on one end it will still bend quite a bit. For example here is an extremely rough FEA where I modeled the whole assembly as a solid block of steel and it moves 0.05mm when 200g of force is applied. Its not awful, but the simulation is probably 2-4 times more rigid than your actual build will be and that kind of deflection will be visible.

a84nVfn.png

The bigger problem IMO is that if you leave the rods unsupported and someone bumps into them you are essentially applying half a meter of leverage to a very small hole in a piece of cast aluminum. I'd guess 2Kg at the top of the rod could permanently damage the alignment of the printer.

Again, not awful, but not really something I'd want to risk.

Your post and edited post are helpful! When you mentioned the 22mm diameter leadscrew as 'doing almost nothing', do you mean that it's just oversized for no purpose other than visually matching the size of the linear shafts?

Would it be possible to lift the XY plate with only one ballscrew and not using my diagonal approach?

In the wood planer I mentioned, it actually uses 4 leadscrews as the posts and uses a closed loop chain to wrap around all four leadnuts:



Edited 4 time(s). Last edit at 03/04/2018 09:15PM by klcjr89.

Looks interesting, maybe consider a frame with linear rails for stability.

Quote
orea
Looks interesting, maybe consider a frame with linear rails for stability.

The whole idea is to use the linear shafts/leadscrews as the frame and have a brand new design of printer. The wood planer has way more forces on it and they use four leadscrews.

Edited 1 time(s). Last edit at 03/04/2018 09:38PM by klcjr89.

The planer doesn't have to worry about stacking 200 um layers of plastic on each other with a few 10s of microns accuracy and precision, and its Z movement range is just a few cm.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
The planer doesn't have to worry about stacking 200 um layers of plastic on each other with a few 10s of microns accuracy and precision, and its Z movement range is just a few cm.

Is that only info you came to provide here? The planer can mantain a very precise Z depth on a piece of wood with loads of cutting force applied.

Quote
klcjr89
Is that only info you came to provide here?

Sometimes pointing out the obvious is the best info to provide.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist

Quote
klcjr89
Is that only info you came to provide here?

Sometimes pointing out the obvious is the best info to provide.

Right! I'm looking for positive feedback like 691175002 and o_lampe provided.

Designing is free, so it's best to solve everything now before ordering expensive bits of kit.

Edited 2 time(s). Last edit at 03/04/2018 10:33PM by klcjr89.

Right now, I'm thinking about converting the design to use two 25mm linear shafts, 'double' length linear bearings, and two 25mm x 5mm pitch ballscrews.

Is it actually possible to use one 25mm ballscrew and three 25mm linear shafts, or would lifting in one corner cause a tilted plane? I was only assuming double length linear bearings would help, but maybe the three corners would 'lag' the driven corner?

Edited 2 time(s). Last edit at 03/04/2018 11:13PM by klcjr89.

I'd go for two ballscrews, but maybe you can drive them by only one stepper? ( belt'n'pulleys)
Don't see the leadscrews as part of the frame. They should only provide z-lift, but no lateral forces. IMHO the nut-bearing mount should be swimming somehow.

Made as sturdy as this wood planer, it can work but what are the advantages compared to a box frame that can be closed to make an enclosure, is sturdier, can use lighter, cheaper parts, guides and lead screw ?
Not to mention the moving gantry instead of the print bed. ?

The box is the most rigid structure for a given weight.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
o_lampe
IMHO the nut-bearing mount should be swimming somehow.

If the ballnut's OD is 40mm, I could solve this with just using a 42mm ID bushing, correct?



Edited 1 time(s). Last edit at 03/05/2018 12:36PM by klcjr89.

The planer is a different application because its only concern is its Z height, whereas a 3d printer needs to be precise in all three axes.

If you have the planer handy you can try pushing it sideways to see how it reacts. I wouldn't be surprised if the wood planer has measurable side-to-side movement, because in use you expect the force to come from only one direction. Another way to think about it is that the nuts have to have clearance (which allows side movement) or they would be unable to rotate freely.

Quote
klcjr89
Your post and edited post are helpful! When you mentioned the 22mm diameter leadscrew as 'doing almost nothing', do you mean that it's just oversized for no purpose other than visually matching the size of the linear shafts?

Pretty much. The linear guides will be more precise than the leadscrew, so they will be doing most of the XY work. The leadscrew is only handling the weight of the positioning stage so you essentially have a pair of 22mm screws holding up 2Kg.

Quote

If the ballnut's OD is 40mm, I could solve this with just using a 42mm ID bushing, correct?

Yes, but floating the nuts is an admission that the screws are only providing support in Z so you don't really need a 25mm ballscrew. Easier to use an 8mm leadscrew and let the screw bend to comply with the linear shafts.

The underlying problem is that you need three points to define a plane. Using two leadscrews and two linear shafts is awkward because you are overconstrained in some ways but underconstrained in others. I'd consider doing this with three or four Chinese 2005 ballscrews and no linear shafts, but using ballscrews as linear guides is rather experimental.

Quote
691175002
The underlying problem is that you need three points to define a plane. Using two leadscrews and two linear shafts is awkward because you are overconstrained in some ways but underconstrained in others. I'd consider doing this with three or four Chinese 2005 ballscrews and no linear shafts, but using ballscrews as linear guides is rather experimental.

If I were to use four ballscrews, how would I go about getting a closed loop belt long enough so I could just use one stepper motor? Four stepper motors would seem to be excessive in weight.

I think three ballscrews would make it awkward to later route the printhead beltpath, and visually might look odd?

I like the feedback and help you are providing

Edited 1 time(s). Last edit at 03/05/2018 01:45PM by klcjr89.

Three screws will be easier to align than four screws but I agree that it would look a little weird. I would probably go with four screws myself.

You can find standard belts in 2m+ lengths if you go up to the ~3mm pitches, but I wouldn't immediately discard the idea of using four steppers. The steppers will only be moving in Z so their weight is essentially irrelevant. Its the moving mass on the gantry or carriage that needs to be minimized.



The primary tradeoff of a seperate stepper for each screw is that you need to keep them synchronized. Automatic leveling will take care of minor misalignments, but if the screws diverge too much the axis could bind.

I'd probably make a call to go the easy route here and rely on automatic bed probing/leveling. Ballscrews can backdrive so it should actually be impossible for the axis to bind. (The screws will return to a relaxed state when unpowered.)

Quote
691175002
Three screws will be easier to align than four screws but I agree that it would look a little weird. I would probably go with four screws myself.

You can find standard belts in 2m+ lengths if you go up to the ~3mm pitches, but I wouldn't immediately discard the idea of using four steppers. The steppers will only be moving in Z so their weight is essentially irrelevant. Its the moving mass on the gantry or carriage that needs to be minimized.

The primary tradeoff of a seperate stepper for each screw is that you need to keep them synchronized. Automatic leveling will take care of minor misalignments, but if the screws diverge too much the axis could bind.

I'd probably make a call to go the easy route here and rely on automatic bed probing/leveling. Ballscrews can backdrive so it should actually be impossible for the axis to bind. (The screws will return to a relaxed state when unpowered.)

If you had the choice, would you go with a single stepper and a closed loop belt? The idea to initially 'level' each corner at first is to lower the XY plate down to where each corner references the bottom Mic-6 plate (use a 123 block or similar), and then at this moment you tighten each pulley to the ballnut with a set screw, keeping synchronization.

I can't wrap my head around how a four stepper motor Z with autoleveling would work.

Interesting concept. Reminds me of the J-Rev
[www.engineering.com]

4 independently driven ballscrews does have the benefit of being able to software leveling.
Tough two rods (or guide rails) on either side would help diminish Z banding, if there was any run out on the screws.

Also, since you mention an enclosure, best if all framing structural elements were steel, just for the sake of thermal expansion.
Yea, and the screws and rods could be mounted to the frame, which could minimize twist.

FWIW, the planar doesn't care about twist, and the screws are all subject to tension (elongation) which steel excels at.

If you spend the money on the controller expansion board, the extra drivers, the cables, the sensors, and the beefier power supply to handle four Z axis motors, you'll end up with a printer that levels the bed automatically and gets that first layer to stick... until something fails, and then you won't be able to print at all.

OTOH, if you use some of the money you would have spent on all that stuff and make a rigid frame for the printer, and drive all the Z axis screws using a single motor, after initial setup the bed will stay level (you won't need autoleveling) and there will be less stuff to fail - i.e. the machine will be more reliable.

If your goal is to make awesome youtube videos, go with the autoleveling. If your goal is to make an awesomely reliable printer, go the other way.

Have you ever sat at a table that has four feet/legs in a cafe or restaurant? Have you ever had to put a piece of cardboard under the table's foot because the table kept rocking? That's what will happen if you use 4 screws to drive the Z axis. 3 points define a plane. Using 3 screws under a planar surface is inherently stable. Adding a fourth adds instability.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
If you spend the money on the controller expansion board, the extra drivers, the cables, the sensors, and the beefier power supply to handle four Z axis motors, you'll end up with a printer that levels the bed automatically and gets that first layer to stick... until something fails, and then you won't be able to print at all.

OTOH, if you use some of the money you would have spent on all that stuff and make a rigid frame for the printer, and drive all the Z axis screws using a single motor, after initial setup the bed will stay level (you won't need autoleveling) and there will be less stuff to fail - i.e. the machine will be more reliable.

If your goal is to make awesome youtube videos, go with the autoleveling. If your goal is to make an awesomely reliable printer, go the other way.

Have you ever sat at a table that has four feet/legs in a cafe or restaurant? Have you ever had to put a piece of cardboard under the table's foot because the table kept rocking? That's what will happen if you use 4 screws to drive the Z axis. 3 points define a plane. Using 3 screws under a planar surface is inherently stable. Adding a fourth adds instability.

In the reliable case of using three synchronized ballscrews with one stepper motor, would you recommend the fourth corner being a 25mm ground linear shaft?

Edited 1 time(s). Last edit at 03/05/2018 04:05PM by klcjr89.

Automatic bed levelling (ABL) and automatic bed compensation (ABC) are completely different things. Sadly, automatic bed compensation is frequently mistakenly called automatic bed levelling. For example, Marlin claims to support "Unified bed levelling", but to the best of my knowledge it does not support bed levelling at all - the page at [marlinfw.org] says plenty about bed compensation but nothing at all about bed leveling.

Both techniques use a Z probe to measure the errors, but that's all they have in common. The difference is this. ABL does what is says: it levels the bed, by adjusting multiple Z stepper motors independently to remove tilt in one or two directions, and possibly also to remove twist. It does not compensate for a bed that isn't flat, or an X gantry that sags. Only a few firmwares support it - AFAIK just RepRapFirmware and Repetier.

ABC is supported by all major 3D printer firmwares. Usually it probes the bed in a grid pattern, then during printing it adjusts the height of the print head relative to the bed to compensate for the observed errors. It can compensate for a bed that isn't flat, and for sag in the X gantry. It can also compensate for small amounts of bed tilt and twist.

If you use multiple Z motors, then when you power your printer off and then on again, the motors may get out of sync. This is because at power up they will jump to the nearest multiple of 4 full steps that matches the motor current, and if they are between such positions, they may jump in different directions. That makes them 4 full steps different from each other compared to last time you used your printer.

You can handle this in a few ways:

1. Ignore it until the difference between Z positions at different points of the bed becomes too great, then resync the motors manually.

2. Use a separate homing switch for each Z motor.

3. Use ABC to compensate (e.g. by doing a mesh bed probe after every power up), until the error is large enough to cause binding. Then resync manually.

4. Use ABL, which resyncs the motors by adjusting them independently.

5. Avoid the problem by using a single motor driving multiple leadscrews using a belt.

IMO the sensible choices are #4 and #5. #4 is mechanically simple but requires a separate stepper driver per Z motor. #5 is mechanically more complex, but electrically simpler.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
klcjr89
If you had the choice, would you go with a single stepper and a closed loop belt? The idea to initially 'level' each corner at first is to lower the XY plate down to where each corner references the bottom Mic-6 plate (use a 123 block or similar), and then at this moment you tighten each pulley to the ballnut with a set screw, keeping synchronization.

That's a hard question to answer, its more of an aesthetic/stylistic choice.

A belt keeps the screws in sync which is very nice; but seperate steppers are more flexible and might even be simpler since you can copy-paste the same assembly for each corner. I'd probably start designing for a belt but fall back to individual steppers fairly quickly if the design becomes too messy.

Quote
klcjr89
I can't wrap my head around how a four stepper motor Z with autoleveling would work.

EDIT: dc42's explanation of your bed leveling/compensation is clearer and more complete than what I wrote, However I'd propose that ballscrews might make #3 plausible as well. It should be impossible for the Z axis to bind because any misaligned screw will backdrive and "spring back" towards alignment when the steppers are off.

Well there are two different ways to use autolevelling here. The simplest way is to connect all the Z motors together and rely on standard grid leveling. So if the XY stage is tilted it will stay tilted, and the entire Z axis will move up and down to compensate. This is conceptually sloppy, but its used on almost every i3 printer and has no material drawback unless an axis binds (which should be impossible when using ballscrews).

If you want the printer to independently manipulate each Z stepper to level the XY stage relative to the bed, you can either use four homing switches (one for each Z stepper) or probe all four corners of the bed then read back those offsets and adjust the appropriate screws. These options are clever, but probably a PITA to implement.

As I mentioned before, using ballscrews as linear guides is discouraged, but the datasheets make it seem pretty plausible. Its entirely possible that screws might be smoother than the Chinese LMU parts. [imgur.com]

Edited 2 time(s). Last edit at 03/05/2018 04:33PM by 691175002.

Quote
691175002
As I mentioned before, using ballscrews as linear guides is discouraged, but the datasheets make it seem pretty plausible. Its entirely possible that screws might be smoother than the Chinese LMU parts. [imgur.com]

Right now I'm leaning on using a single motor with four ballscrews. Two questions: are 20mm ballscrews enough in diameter, why not 25mm? Lastly, Misumi has C10 ballscrews for reduced cost, do you think these would be plenty good enough?

The plan for the printer (subject to change) is 400mm^3 build volume, with DyzEnd hotend and liquid cooling.

The linear guides I plan to use for the printhead are LM76 speed demon, size 10 with 3 rollers (two fixed + one eccentric), unless you recommend against them?



Edited 6 time(s). Last edit at 03/05/2018 04:43PM by klcjr89.

Virtually any ballscrew should be fine here. I remember the Chinese rolled screws being roughly a third of the price of Misumi's C10 screws, and have found them to be acceptable, so you could give that some consideration.

Appropriate screw diameter depends on how they are being supported. The weak point is likely to be the connection between the base and the screw, so unless you have a good plan for that interface larger screws won't help much.

I'm not a huge fan of pressing the screws into 1/2" plate because of the leverage it creates. I would probably be looking at putting a huge steel washer/shim on one or both sides of the plate to distribute the load, and machining the end of the ballscrew so that it clamps down on the plate + washer/shim.

[us.misumi-ec.com]

Quote
691175002
Virtually any ballscrew should be fine here. I remember the Chinese rolled screws being roughly a third of the price of Misumi's C10 screws, and have found them to be acceptable, so you could give that some consideration.

Appropriate screw diameter depends on how they are being supported. The weak point is likely to be the connection between the base and the screw, so unless you have a good plan for that interface larger screws won't help much.

I'm not a huge fan of pressing the screws into 1/2" plate because of the leverage it creates. I would probably be looking at putting a huge steel washer/shim on one or both sides of the plate to distribute the load, and machining the end of the ballscrew so that it clamps down on the plate + washer/shim.

[us.misumi-ec.com]

My concern with smaller ballscrews is they will be less straight than their larger counterparts, whether from shipping damage or just due to the fact that the cross section is thinner.

In regards to press fitting, would it be ok to do so if there is a top plate press fitted on the top of the screw ends, and then have side plates that secure the printer's bottom and top plates together? The XY plate would have to be inserted first of course, but this is ok.

Do you think those speed demon linear guides would work well?

Quote
klcjr89
In the reliable case of using three synchronized ballscrews with one stepper motor, would you recommend the fourth corner being a 25mm ground linear shaft?

I would not recommend that configuration at all. Screws provide motive force. Guide rails provide guidance. One round guide rail allows the bed to go up and down along the axis of the rail, and allows it to rotate around the rail. You need a second rail to prevent the rotation. 3 screws and 2 guide rails should do the job. I would not try to use screws or guide rails as structural elements.

I know you like the way your design looks, but it isn't very good mechanically. I think you have to decide if you want your printer to look nice or perform well. If you want both you may have to explore options that allow for more than 4 vertical elements.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]